Re: [PATCH 1/1] ntb: use dmaengine_get_dma_device() instead of chan->device->dev

[Frank Li <Frank.li@oss.nxp.com>]

On Fri, Sep 18, 2026 at 08:53:23PM +0000, sashiko-bot@kernel.org wrote:
> Thank you for your contribution! Sashiko AI review found 2 potential issue(s) to consider:
> - [Medium] ntb: missed replacing chan->device->dev in perf_dma_filter()
> - [Low] ntb: commit message incorrectly mentions IIO instead of NTB
> --
>
> commit 00ad86ab4ff3c677b79412d332638c7a5186dbbe
> Author: Frank Li <Frank.Li@nxp.com>
>
> ntb: use dmaengine_get_dma_device() instead of chan->device->dev
>
> This commit replaces direct chan->device->dev access with the proper
> dmaengine_get_dma_device() accessor in NTB DMA consumers. It prepares
> for making the DMA engine provider data structures private.
>
> >     Replace direct dma_chan::device::dev access with the proper
> >     dmaengine_get_dma_device() accessor in IIO DMA consumers.
>
> [Severity: Low]
> This isn't a bug, but the commit message incorrectly mentions IIO DMA
> consumers instead of NTB DMA consumers.
>
> > diff --git a/drivers/ntb/test/ntb_perf.c b/drivers/ntb/test/ntb_perf.c
> > index dfd175f79e8f0..b936f15079df2 100644
> > --- a/drivers/ntb/test/ntb_perf.c
> > +++ b/drivers/ntb/test/ntb_perf.c
> [ ... ]
> > @@ -869,6 +869,7 @@ static bool perf_dma_filter(struct dma_chan *chan, void *data)
> >  static int perf_init_test(struct perf_thread *pthr)
> >  {
> >  	struct perf_ctx *perf = pthr->perf;
> > +	struct device *dma_dev;
>
> [Severity: Medium]
> Does this patch miss a replacement in perf_dma_filter()?
>
> Right above this hunk, perf_dma_filter() still accesses chan->device->dev
> directly:
>
> drivers/ntb/test/ntb_perf.c:perf_dma_filter() {
>     ...
>     return node == NUMA_NO_NODE || node == dev_to_node(chan->device->dev);
> }

Can't simple replace it here. Following patch will handle it by add
new DMA Engine API.

which require dma provider device to get dma channel. if return per channel
device, will fail to get dma chan.

Frank

>
> Since the goal is to prepare for making DMA engine structures private by
> abstracting chan->device->dev behind dmaengine_get_dma_device(chan), will
> leaving this direct access cause a build failure once the structure is made
> private?
>
> --
> Sashiko AI review · https://sashiko.dev/#/patchset/20260918204152.2364336-1-Frank.Li@oss.nxp.com?part=1